Daniel Martin Katz (Chicago Kent), Michael James Bommarito (Michigan State), Shang Gao (Casetext) & Pablo Arredondo (Casetext), GPT-4 Passes the Bar Examination:

On this paper, we experimentally consider the zero-shot efficiency of a preliminary model of GPT-4 towards prior generations of GPT on your complete Uniform Bar Examination (UBE), together with not solely the multiple-choice Multistate Bar Examination (MBE), but in addition the open-ended Multistate Essay Examination (MEE) and Multistate Efficiency Check (MPT) elements. On the MBE, GPT-4 considerably outperforms each human test-takers and prior fashions, demonstrating a 26% enhance over ChatGPT and beating people in 5 of seven topic areas. On the MEE and MPT, which haven’t beforehand been evaluated by students, GPT-4 scores a median of 4.2/6.0 as in comparison with a lot decrease scores for ChatGPT. Graded throughout the UBE elements, within the method during which a human tast-taker can be, GPT-4 scores roughly 297 factors, considerably in extra of the passing threshold for all UBE jurisdictions. These findings doc not simply the fast and noteworthy advance of enormous language mannequin efficiency typically, but in addition the potential for such fashions to help the supply of authorized companies in society.

• ABA Journal, Newest Model of ChatGPT Aces Bar Examination With Rating Nearing ninetieth Percentile
• Josh Blackman (South Texas), In January, ChatGPT Failed The Bar. In March, GPT-4 Exceeds The Nationwide Pupil Common
• Casetext Press Launch, CoCounsel Builds on the Energy of GPT-4, the AI That Outperformed Actual Bar Candidates
• Chicago-Kent Press Launch, GPT-4 Passes the Bar Examination
• Reuters, Bar Examination Rating Reveals AI Can Hold Up With ‘Human Legal professionals,’ Researchers Say
• Eugene Volokh (UCLA), ChatGPT-4 Sharply Outperforms ChatGPT-3. However People-4 Do not Sharply Outperform People-3

https://taxprof.typepad.com/taxprof_blog/2023/03/gpt-4-beats-90-of-aspiring-lawyers-on-the-bar-exam.html